Excluding the Supernatural

Oc­ca­sion­ally, you hear some­one claiming that cre­ation­ism should not be taught in schools, es­pe­cially not as a com­pet­ing hy­poth­e­sis to evolu­tion, be­cause cre­ation­ism is a pri­ori and au­to­mat­i­cally ex­cluded from sci­en­tific con­sid­er­a­tion, in that it in­vokes the “su­per­nat­u­ral”.

So… is the idea here, that cre­ation­ism could be true, but even if it were true, you wouldn’t be al­lowed to teach it in sci­ence class, be­cause sci­ence is only about “nat­u­ral” things?

It seems clear enough that this no­tion stems from the de­sire to avoid a con­fronta­tion be­tween sci­ence and re­li­gion. You don’t want to come right out and say that sci­ence doesn’t teach Reli­gious Claim X be­cause X has been tested by the sci­en­tific method and found false. So in­stead, you can… um… claim that sci­ence is ex­clud­ing hy­poth­e­sis X a pri­ori. That way you don’t have to dis­cuss how ex­per­i­ment has falsified X a pos­te­ri­ori.

Of course this plays right into the cre­ation­ist claim that In­tel­li­gent De­sign isn’t get­ting a fair shake from sci­ence—that sci­ence has pre­judged the is­sue in fa­vor of athe­ism, re­gard­less of the ev­i­dence. If sci­ence ex­cluded In­tel­li­gent De­sign a pri­ori, this would be a jus­tified com­plaint!

But let’s back up a mo­ment. The one comes to you and says: “In­tel­li­gent De­sign is ex­cluded from be­ing sci­ence a pri­ori, be­cause it is ‘su­per­nat­u­ral’, and sci­ence only deals in ‘nat­u­ral’ ex­pla­na­tions.”

What ex­actly do they mean, “su­per­nat­u­ral”? Is any ex­pla­na­tion in­vented by some­one with the last name “Co­hen” a su­per­nat­u­ral one? If we’re go­ing to sum­mar­ily kick a set of hy­pothe­ses out of sci­ence, what is it that we’re sup­posed to ex­clude?

By far the best defi­ni­tion I’ve ever heard of the su­per­nat­u­ral is Richard Car­rier’s: A “su­per­nat­u­ral” ex­pla­na­tion ap­peals to on­tolog­i­cally ba­sic men­tal things, men­tal en­tities that can­not be re­duced to non­men­tal en­tities.

This is the differ­ence, for ex­am­ple, be­tween say­ing that wa­ter rolls down­hill be­cause it wants to be lower, and set­ting forth differ­en­tial equa­tions that claim to de­scribe only mo­tions, not de­sires. It’s the differ­ence be­tween say­ing that a tree puts forth leaves be­cause of a tree spirit, ver­sus ex­am­in­ing plant bio­chem­istry. Cog­ni­tive sci­ence takes the fight against su­per­nat­u­ral­ism into the realm of the mind.

Why is this an ex­cel­lent defi­ni­tion of the su­per­nat­u­ral? I re­fer you to Richard Car­rier for the full ar­gu­ment. But con­sider: Sup­pose that you dis­cover what seems to be a spirit, in­hab­it­ing a tree: a dryad who can ma­te­ri­al­ize out­side or in­side the tree, who speaks in English about the need to pro­tect her tree, et cetera. And then sup­pose that we turn a micro­scope on this tree spirit, and she turns out to be made of parts—not in­her­ently spiritual and in­ef­fable parts, like fabric of de­sire­ness and cloth of be­lief; but rather the same sort of parts as quarks and elec­trons, parts whose be­hav­ior is defined in mo­tions rather than minds. Wouldn’t the dryad im­me­di­ately be de­moted to the dull cat­a­logue of com­mon things?

But if we ac­cept Richard Car­rier’s defi­ni­tion of the su­per­nat­u­ral, then a dilemma arises: we want to give re­li­gious claims a fair shake, but it seems that we have very good grounds for ex­clud­ing su­per­nat­u­ral ex­pla­na­tions a pri­ori.

I mean, what would the uni­verse look like if re­duc­tion­ism were false?

I pre­vi­ously defined the re­duc­tion­ist the­sis as fol­lows: hu­man minds cre­ate multi-level mod­els of re­al­ity in which high-level pat­terns and low-level pat­terns are sep­a­rately and ex­plic­itly rep­re­sented. A physi­cist knows New­ton’s equa­tion for grav­ity, Ein­stein’s equa­tion for grav­ity, and the deriva­tion of the former as a low-speed ap­prox­i­ma­tion of the lat­ter. But these three sep­a­rate men­tal rep­re­sen­ta­tions, are only a con­ve­nience of hu­man cog­ni­tion. It is not that re­al­ity it­self has an Ein­stein equa­tion that gov­erns at high speeds, a New­ton equa­tion that gov­erns at low speeds, and a “bridg­ing law” that smooths the in­ter­face. Real­ity it­self has only a sin­gle level, Ein­stei­nian grav­ity. It is only the Mind Pro­jec­tion Fal­lacy that makes some peo­ple talk as if the higher lev­els could have a sep­a­rate ex­is­tence—differ­ent lev­els of or­ga­ni­za­tion can have sep­a­rate rep­re­sen­ta­tions in hu­man maps, but the ter­ri­tory it­self is a sin­gle unified low-level math­e­mat­i­cal ob­ject.

Sup­pose this were wrong.

Sup­pose that the Mind Pro­jec­tion Fal­lacy was not a fal­lacy, but sim­ply true.

Sup­pose that a 747 had a fun­da­men­tal phys­i­cal ex­is­tence apart from the quarks mak­ing up the 747.

What ex­per­i­men­tal ob­ser­va­tions would you ex­pect to make, if you found your­self in such a uni­verse?

If you can’t come up with a good an­swer to that, it’s not ob­ser­va­tion that’s rul­ing out “non-re­duc­tion­ist” be­liefs, but a pri­ori log­i­cal in­co­her­ence. If you can’t say what pre­dic­tions the “non-re­duc­tion­ist” model makes, how can you say that ex­per­i­men­tal ev­i­dence rules it out?

My the­sis is that non-re­duc­tion­ism is a con­fu­sion; and once you re­al­ize that an idea is a con­fu­sion, it be­comes a tad difficult to en­vi­sion what the uni­verse would look like if the con­fu­sion were true. Maybe I’ve got some multi-level model of the world, and the multi-level model has a one-to-one di­rect cor­re­spon­dence with the causal el­e­ments of the physics? But once all the rules are speci­fied, why wouldn’t the model just flat­ten out into yet an­other list of fun­da­men­tal things and their in­ter­ac­tions? Does ev­ery­thing I can see in the model, like a 747 or a hu­man mind, have to be­come a sep­a­rate real thing? But what if I see a pat­tern in that new su­per­sys­tem?

Su­per­nat­u­ral­ism is a spe­cial case of non-re­duc­tion­ism, where it is not 747s that are ir­re­ducible, but just (some) men­tal things. Reli­gion is a spe­cial case of su­per­nat­u­ral­ism, where the ir­re­ducible men­tal things are God(s) and souls; and per­haps also sins, an­gels, karma, etc.

If I pro­pose the ex­is­tence of a pow­er­ful en­tity with the abil­ity to sur­vey and al­ter each el­e­ment of our ob­served uni­verse, but with the en­tity re­ducible to non­men­tal parts that in­ter­act with the el­e­ments of our uni­verse in a lawful way; if I pro­pose that this en­tity wants cer­tain par­tic­u­lar things, but “wants” us­ing a brain com­posed of par­ti­cles and fields; then this is not yet a re­li­gion, just a nat­u­ral­is­tic hy­poth­e­sis about a nat­u­ral­is­tic Ma­trix. If to­mor­row the clouds parted and a vast glow­ing amor­phous figure thun­dered forth the above de­scrip­tion of re­al­ity, then this would not im­ply that the figure was nec­es­sar­ily hon­est; but I would show the movies in a sci­ence class, and I would try to de­rive testable pre­dic­tions from the the­ory.

Con­versely, re­li­gions have ig­nored the dis­cov­ery of that an­cient bod­iless thing: om­nip­re­sent in the work­ing of Na­ture and im­ma­nent in ev­ery fal­ling leaf: vast as a planet’s sur­face and billions of years old: it­self un­made and aris­ing from the struc­ture of physics: de­sign­ing with­out brain to shape all life on Earth and the minds of hu­man­ity. Nat­u­ral se­lec­tion, when Dar­win pro­posed it, was not hailed as the long-awaited Creator: It wasn’t fun­da­men­tally men­tal.

But now we get to the dilemma: if the staid con­ven­tional nor­mal bor­ing un­der­stand­ing of physics and the brain is cor­rect, there’s no way in prin­ci­ple that a hu­man be­ing can con­cretely en­vi­sion, and de­rive testable ex­per­i­men­tal pre­dic­tions about, an al­ter­nate uni­verse in which things are ir­re­ducibly men­tal. Be­cause, if the bor­ing old nor­mal model is cor­rect, your brain is made of quarks, and so your brain will only be able to en­vi­sion and con­cretely pre­dict things that can pre­dicted by quarks. You will only ever be able to con­struct mod­els made of in­ter­act­ing sim­ple things.

Peo­ple who live in re­duc­tion­ist uni­verses can­not con­cretely en­vi­sion non-re­duc­tion­ist uni­verses. They can pro­nounce the syl­la­bles “non-re­duc­tion­ist” but they can’t imag­ine it.

The ba­sic er­ror of an­thro­po­mor­phism, and the rea­son why su­per­nat­u­ral ex­pla­na­tions sound much sim­pler than they re­ally are, is your brain us­ing it­self as an opaque black box to pre­dict other things la­beled “mind­ful”. Be­cause you already have big, com­pli­cated webs of neu­ral cir­cuitry that im­ple­ment your “want­ing” things, it seems like you can eas­ily de­scribe wa­ter that “wants” to flow down­hill—the one word “want” acts as a lever to set your own com­pli­cated want­ing-ma­chin­ery in mo­tion.

Or you imag­ine that God likes beau­tiful things, and there­fore made the flow­ers. Your own “beauty” cir­cuitry de­ter­mines what is “beau­tiful” and “not beau­tiful”. But you don’t know the di­a­gram of your own synapses. You can’t de­scribe a non­men­tal sys­tem that com­putes the same la­bel for what is “beau­tiful” or “not beau­tiful”—can’t write a com­puter pro­gram that pre­dicts your own la­bel­ings. But this is just a defect of knowl­edge on your part; it doesn’t mean that the brain has no ex­pla­na­tion.

If the “bor­ing view” of re­al­ity is cor­rect, then you can never pre­dict any­thing ir­re­ducible be­cause you are re­ducible. You can never get Bayesian con­fir­ma­tion for a hy­poth­e­sis of ir­re­ducibil­ity, be­cause any pre­dic­tion you can make is, there­fore, some­thing that could also be pre­dicted by a re­ducible thing, namely your brain.

Some boxes you re­ally can’t think out­side. If our uni­verse re­ally is Tur­ing com­putable, we will never be able to con­cretely en­vi­sion any­thing that isn’t Tur­ing-com­putable—no mat­ter how many lev­els of halt­ing or­a­cle hi­er­ar­chy our math­e­mat­i­ci­ans can talk about, we won’t be able to pre­dict what a halt­ing or­a­cle would ac­tu­ally say, in such fash­ion as to ex­per­i­men­tally dis­crim­i­nate it from merely com­putable rea­son­ing.

Of course, that’s all as­sum­ing the “bor­ing view” is cor­rect. To the ex­tent that you be­lieve evolu­tion is true, you should not ex­pect to en­counter strong ev­i­dence against evolu­tion. To the ex­tent you be­lieve re­duc­tion­ism is true, you should ex­pect non-re­duc­tion­ist hy­pothe­ses to be in­co­her­ent as well as wrong. To the ex­tent you be­lieve su­per­nat­u­ral­ism is false, you should ex­pect it to be in­con­ceiv­able as well.

If, on the other hand, a su­per­nat­u­ral hy­poth­e­sis turns out to be true, then pre­sum­ably you will also dis­cover that it is not in­con­ceiv­able.

So let us bring this back full cir­cle to the mat­ter of In­tel­li­gent De­sign:

Should ID be ex­cluded a pri­ori from ex­per­i­men­tal falsifi­ca­tion and sci­ence class­rooms, be­cause, by in­vok­ing the su­per­nat­u­ral, it has placed it­self out­side of nat­u­ral philos­o­phy?

I an­swer: “Of course not.” The ir­re­ducibil­ity of the in­tel­li­gent de­signer is not an in­dis­pens­able part of the ID hy­poth­e­sis. For ev­ery ir­re­ducible God that can be pro­posed by the IDers, there ex­ists a cor­re­spond­ing re­ducible alien that be­haves in ac­cor­dance with the same pre­dic­tions—since the IDers them­selves are re­ducible; to the ex­tent I be­lieve re­duc­tion­ism is in fact cor­rect, which is a rather strong ex­tent, I must ex­pect to dis­cover re­ducible for­mu­la­tions of all sup­pos­edly su­per­nat­u­ral pre­dic­tive mod­els.

If we’re go­ing over the arche­olog­i­cal records to test the as­ser­tion that Je­ho­vah parted the Red Sea out of an ex­plicit de­sire to dis­play its su­per­hu­man power, then it makes lit­tle differ­ence whether Je­ho­vah is on­tolog­i­cally ba­sic, or an alien with nan­otech, or a Dark Lord of the Ma­trix. You do some arche­ol­ogy, find no skele­tal rem­nants or ar­mor at the Red Sea site, and in­deed find records that Egypt ruled much of Canaan at the time. So you stamp the his­tor­i­cal record in the Bible “dis­proven” and carry on. The hy­poth­e­sis is co­her­ent, falsifi­able and wrong.

Like­wise with the ev­i­dence from biol­ogy that foxes are de­signed to chase rab­bits, rab­bits are de­signed to evade foxes, and nei­ther is de­signed “to carry on their species” or “pro­tect the har­mony of Na­ture”; like­wise with the retina be­ing de­signed back­wards with the light-sen­si­tive parts at the bot­tom; and so on through a thou­sand other items of ev­i­dence for splin­tered, im­moral, in­com­pe­tent de­sign. The Je­ho­vah model of our alien god is co­her­ent, falsifi­able, and wrong—co­her­ent, that is, so long as you don’t care whether Je­ho­vah is on­tolog­i­cally ba­sic or just an alien.

Just con­vert the su­per­nat­u­ral hy­poth­e­sis into the cor­re­spond­ing nat­u­ral hy­poth­e­sis. Just make the same pre­dic­tions the same way, with­out as­sert­ing any men­tal things to be on­tolog­i­cally ba­sic. Con­sult your brain’s black box if nec­es­sary to make pre­dic­tions—say, if you want to talk about an “an­gry god” with­out build­ing a full-fledged an­gry AI to la­bel be­hav­iors as an­gry or not an­gry. So you de­rive the pre­dic­tions, or look up the pre­dic­tions made by an­cient the­olo­gians with­out ad­vance knowl­edge of our ex­per­i­men­tal re­sults. If ex­per­i­ment con­flicts with those pre­dic­tions, then it is fair to speak of the re­li­gious claim hav­ing been sci­en­tifi­cally re­futed. It was given its just chance at con­fir­ma­tion; it is be­ing ex­cluded a pos­te­ri­ori, not a pri­ori.

Ul­ti­mately, re­duc­tion­ism is just dis­be­lief in fun­da­men­tally com­pli­cated things. If “fun­da­men­tally com­pli­cated” sounds like an oxy­moron… well, that’s why I think that the doc­trine of non-re­duc­tion­ism is a con­fu­sion, rather than a way that things could be, but aren’t. You would be wise to be wary, if you find your­self sup­pos­ing such things.

But the ul­ti­mate rule of sci­ence is to look and see. If ever a God ap­peared to thun­der upon the moun­tains, it would be some­thing that peo­ple looked at and saw.

Corol­lary: Any sup­posed de­signer of Ar­tifi­cial Gen­eral In­tel­li­gence who talks about re­li­gious be­liefs in re­spect­ful tones, is clearly not an ex­pert on re­duc­ing men­tal things to non­men­tal things; and in­deed knows so very lit­tle of the ut­ter­most ba­sics, as for it to be scarcely plau­si­ble that they could be ex­pert at the art; un­less their idiot sa­vancy is com­plete. Or, of course, if they’re out­right ly­ing. We’re not talk­ing about a sub­tle mis­take.